Photodiodes are widely used for sensing light radiation. There are many applications in which the level of the light which is required to be sensed is very low, and therefore the sensitivity of said photodiodes is a critical requirement.
It is well known in the art that the signal-to-noise ratio which can be obtained from photodiodes (and from many other electronic components) is limited by the level of the “thermal noise”, which in turn is related to the temperature of the component. The term “dark current” is commonly used in the art to define the current flowing in a photodiode during a total dark condition. The signal-to-noise ratio in photodiodes is conventionally improved by cooling the component, in many cases down to very low temperatures close to 0° K. The means for cooling and maintaining such a low temperature in photodiodes, however, are cumbersome and expensive, and in any case can reduce the noise down to a limited value.
The dark current is generally composed of two main components. The first component, hereinafter referred to as “the diffusion dark current” is due to the thermal excitation of carriers across the complete energy bandgap of the photodiode material. As said, the level of this current can be reduced by means of cooling the component. The second component affecting the level of the dark current is known as the “Generation-Recombination” current hereinafter “G-R dark current”). The level of the G-R dark current can also be reduced by cooling, but at a slower rate of reduction with temperature.
At low temperatures, where the level of the diffusion dark current is reduced sufficiently, the G-R dark current generally becomes the most dominant component of the dark current. There have been made many efforts in trying to reduce the level of the thermal noise. However, there are not known many of such efforts for reducing the G-R current.
FIG. 1 is a band diagram showing the principle of operation of a photodiode according to the prior art. In a semiconductor p-n junction 1-2, a depletion region 3 is formed around the metallurgical junction due to the transfer of electrons from donors in the n-side 2 of the depletion region to acceptors in the p-side 1. The conduction band (EC) and valence band (EV) are bent in the depletion region. This bending is associated with an electric field that drives electrons 7 towards the n-side and holes 8 towards the p-side of the junction. When a bias is applied to the junction, quasi Fermi levels can be defined in each of the two “flat-band” regions. The quasi Fermi level lies near the valence band on the p-side (EF(p)) and near the conduction band on the n-side (EF(n)). At zero bias, the energies of the two quasi Fermi levels are equal. The energy separation of the two quasi Fermi levels in electron-volts is equal to the applied bias in volts. If a reverse bias Vrev is applied to the diode, the following relationship holds:Vrev=EF(p)−EF(n).
The energy gap is given by EG=EC−EV. Although EC and EV change with position due to the band bending in the depletion region, their energy separation is constant everywhere for a “homo-junction” diode (“homo-junction” means that the same material is used on each side of the p-n junction).
Light 9 can be absorbed by promoting an electron 119 from the valence band to the conduction band. The missing electron in the valence band is called a hole, and is indicated by numeral 118. The longest wavelength for this process is called the cut off wavelength and is given by: λC=hc/EG, wherein h is Planck's constant and c is the velocity of light.
The “photo-created” hole 118 in process 9 exists in the n-type material 2 and so is a minority carrier. It can diffuse, as indicated by numeral 10 to the depletion region where it is accelerated 8 into the p-side 1 by the electric field in the depletion region 3. An analogous process can occur in the p-type material 1 where a minority electron is created by the absorption of light. It can diffuse to the depletion region where it is accelerated 7 into the n-side 2 by the electric field in the depletion region 3.
Generation-Recombination (G-R) centers 4, also known as Shockley-Read traps or Shockley-Hall-Read traps, are energy levels that lie close to the middle of the band gap. They are related to imperfections or impurities inside the crystal. The probability of process 9 to occur due to heat (in the absence of an external photon flux) is essentially proportional to exp(−EG/kT) where k is Boltzman's constant and T is the absolute temperature. This process (and the equivalent process on the p-side) gives rise to the “dark current” in a perfect diode with no G-R centers. In this case the dark current is all due to diffusion dark current, and the device is said to be at “the diffusion limit”.
In an asymmetric p+-n homo-junction, where the p-doping is several orders of magnitude greater than the n-doping, it can easily be shown that, in the diffusion limit, the higher of the two minority carrier concentrations, in said p+-n case the minority holes on the n-side, makes the dominant contribution to the dark current.
Since free electrons 7 and holes 8 are removed efficiently by the electric field in the depletion region 3, especially when a reverse bias is applied, an electron that undergoes excitation 5 from the valence band EV to the G-R center 4 cannot return to the valence band. It can only be further excited 6 to the conduction band. Processes 5, 6, 7, and 8 thus give rise to the G-R dark current.
The rate of electron generation by traps, in unit volume of the reverse biased depletion region 3 due to a process, 5, 6, 7, and 8, is approximately described by the formula
                    G        =                              n            1            2                                                              τ                                  n                  ⁢                                                                          ⁢                  0                                            ⁢                              p                ′                                      +                                          τ                                  p                  ⁢                                                                          ⁢                  0                                            ⁢                              n                ′                                                                        (        1        )            where ni is the so called intrinsic carrier concentration (the carrier concentration in the perfectly pure material) and τn0, τp0 are the electron and hole minority carrier lifetimes. This formula may be found, for example, as equation (8.9.2) in chapter 8 of the book by Shyh Wang, entitled “Fundamentals of Semiconductor Theory and Device Physics” (published by Prentice Hall, ISBN 0-13-344425-2). Here n′=n.e(Et−EF)/kT and p′=p.e(EF−Et)/kT where n, p, and EF are the electron concentration, the hole concentration and the Fermi level respectively in a given sample of the semiconductor material, Et is the energy of the trap, and T is the absolute temperature. It can be demonstrated that G in equation (1) is largest when the trap lies near the middle of the energy bandgap. In this case it is easy to show using the above formulae, that
                    G        ≈                              n            1                                (                                          τ                                  n                  ⁢                                                                          ⁢                  0                                            +                              τ                                  p                  ⁢                                                                          ⁢                  0                                                      )                                              (        2        )            
Hence it follows that G is proportional to the intrinsic carrier concentration, the formula for which contains an exponential factor: exp(−EG/2kT). The dark current due to generation-recombination centers is itself proportional to G and so will also vary essentially as: exp(−EG/2kT). It is the weaker temperature dependence of the G-R contribution to the dark current (exp(−EG/2kT)) compared with the diffusion contribution (exp(−EG/kT)) that causes the G-R contribution to dominate at low temperatures. The ratio of the G-R dark current to the diffusion dark current in a p+-n diode is given by equation (8.9.6) in chapter 8 of the earlier mentioned book by Shyh Wang, as:
                                          J                          G              -              R                                            J            diff                          =                                            L              dep                                      L              p                                ×                                    N              D                                      n              ′                                                          (        3        )            where Ldep is the thickness of the depletion region, and ND and Lp are the doping and minority carrier diffusion length on the n-side of the junction. Typical values of Ldep and Lp are ˜0.5μ and 20μ respectively.
Typical narrow gap homo-junction photo-diodes based on e.g. InSb, InAsSb, HgCdTe, etc., are in many cases operated at reduced temperatures, in order to limit the dark current. For such devices operated at 77K, G-R centers typically increase the dark current above the diffusion limit by at least 3-4 orders of magnitude in the MWIR (3-5μ) and 1-2 orders of magnitude in the LWIR (8-12μ) cut-off wavelength regions, behaviour that in each case is consistent with equation (3). This effect may easily be seen in J Bajaj, SPIE proceedings no. 3948 page 45 (FIG. 3 of this article), San Jose, January 2000, or in P. C. Klipstein et al., SPIE proceedings number 4820, page 653 (FIG. 2 of this article), Seattle, July 2002.
The prior art has failed to specifically address the issue of suppressing the G-R contribution to the current by a suitable hetero-junction design.
A design published by J. L. Johnson et al., Journal of Applied Physics, volume 80, pages 1116-1127 (FIG. 3) shows a diode made between an n-type narrow bandgap semiconductor with a relatively low doping level and formed from a type II InAs/Ga1-xInxSb superlattice, and a p-type wide bandgap semiconductor with a relatively high doping level, formed from GaSb. This asymmetric doping ensures that most of the depletion region, with its associated electric field, exists in the narrow bandgap photon absorbing layer made from the type II superlattice. There is no discussion in the article about the importance of removing the electric field from this narrow bandgap region. It appears that the main reason for using a heterojunction p-contact instead of a homojunction p-contact in this article is one of convenience, since the p-type heterojunction contact is easier to grow than a p-type type II superlattice.
FIG. 2 of the article by C. T. Elliott “Advanced Heterostructures for In1-xAlxSb and Hg1-xCdxTe detectors and emitters”, SPIE proceedings vol. 2744, page 452, discloses photodiode devices in which the dark current is reduced by means of the suppression of Auger-related generation processes. Hereinafter, these devices will be referred to shortly as “Elliott devices”. In contrast to the present invention, whose essential part, as will be shown hereinafter, has a wide bandgap semiconductor sandwiched between n-type and p-type semiconductors with similar or narrower bandgaps, the essential part of said Elliott devices has a narrow bandgap semiconductor, clad on each side by an n-type and a p-type semiconductor respectively, each with a larger effective bandgap.
As will be further shown hereinafter, the Elliott devices are based on a different principle than that of the present invention. They are aimed for operating essentially at higher temperatures than for the devices of the present invention, typically room temperature or slightly cooler, in which thermal generation across the bandgap is significant. Under this condition, Auger processes are known to limit drastically the carrier lifetime. By applying a sufficiently large reverse bias to an Elliott device, the free carrier concentration may be reduced to a level characteristic of a lower temperature, so that the Auger processes are suppressed, and the reverse bias dark current or “saturation current” is reduced.
In the article “Advanced Heterostructures for In1-xAlxSb and Hg1-xCdxTe detectors and emitters” by C. T. Elliott, SPIE proceedings vol. 2744, pages 452-462, it is stated (page 453): “Minority carrier exclusion and extraction occur at the pπ and πn junctions respectively and the densities of both carrier types in the active π region decrease . . . as a consequence the thermal generation rates involving Auger processes fall, so that the saturation leakage current is less than would be expected from the zero bias resistance and a region of negative conductance is predicted to occur”. The article then goes on to point out that, in contrast to the object of the present invention, G-R currents are not suppressed. It states: “In InSb devices with a π active region, however, the density and energy of Shockley-Read traps is such that an increase in thermal generation through traps occurs as the diodes are reverse biased, so that negative conductance is only observed above room temperature”. From this statement it may be learned that even at room temperature, an Elliott device based on InSb, exhibits large G-R currents in reverse bias. This is to be expected because there is a significant depletion layer, with an associated electric field, in the low doped π-region of the device, which is also the region with the narrowest bandgap.
There are several other embodiments of the Elliott device based on other materials. For example in the article by A. Rakovska, V. Berger, X. Marcadet, B. Vinter, G. Glastre, in Applied Physics Letters, volume 77, page 397 (2000), a device is described with a photon absorbing layer of InAs0.91Sb0.09. In this case, diffusion limited behaviour was observed down to 200K, as expected at high operating temperatures. At lower temperatures, where the G-R dark current might be expected to dominate, leakage currents dominated instead due to the lack of a suitable surface passivation treatment. The authors speculate in their conclusion that by increasing the bandgap of one of the cladding layers they might be able to further reduce the diffusion dark current above 200K to the point where the G-R dark current is dominant. The clear implication is that since the diffusion dark current reduces faster with temperature, the G-R dark current is expected to dominate below 200K and no special steps are taken to avoid this.
It is an object of the present invention to provide a photodiode in which the dark current is significantly reduced, particularly at low temperatures, generally in the range of about 77 to 200° K, depending on the material and wavelength of operation.
It is a particular object of the present invention to provide a photodiode in which the level of the G-R current is significantly suppressed in a given temperature.
It is still an object of the present invention to reduce the need for cooling, by providing a photodiode structure having a level of dark current that would alternatively exist in a much lower temperature.
It is still a further object of the invention to provide a method and process for manufacturing the photodiode of the present invention.
Other objects and advantages of the present invention will become apparent as the description proceeds.